• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

3070 Ti - Big performance issues after driver update

haste18

New Member
Joined
Mar 19, 2024
Messages
12 (0.03/day)
I’ve been having issues with my laptop performance. It’s a I9-12900H with a 3070 Ti card on Windows 10. Was running Cyberpunk 2077 fine, but noticed FPS dropping. Updated GPU drivers and after that it only got worse. Max 8 FPS and CPU often gets stuck at 0.39GHz. Only way to get it above is by unplugging the PSU and plug it back in. I’ve read many threads on Reddit etc, but nothing really fixes the issue.

This is not the first time I’ve had these issues. It’s been going on for months. A fresh Windows install made the issues disappear, but Windows update and driver updates make it all come back.

I’m new with GPU-Z and looking for guidance with the readings. Can anybody point me in the right direction? Thanks in advance.

1st screenshot shows GPU-Z idle
1710844133291.gif


2nd screenshot shows GPU-Z 40 seconds into OCCT stability power test (CPU + GPU:3D)
1710844163725.gif


3rd screenshot OCCT stability power test
1710844190866.png
 
PerfCap reason THRM at surprisingly low clock and power.

Downgrade the driver and see if the problem goes away
 
I’ve been having issues with my laptop performance. It’s a I9-12900H with a 3070 Ti card on Windows 10. Was running Cyberpunk 2077 fine, but noticed FPS dropping. Updated GPU drivers and after that it only got worse. Max 8 FPS and CPU often gets stuck at 0.39GHz. Only way to get it above is by unplugging the PSU and plug it back in. I’ve read many threads on Reddit etc, but nothing really fixes the issue.

This is not the first time I’ve had these issues. It’s been going on for months. A fresh Windows install made the issues disappear, but Windows update and driver updates make it all come back.

I’m new with GPU-Z and looking for guidance with the readings. Can anybody point me in the right direction? Thanks in advance.

1st screenshot shows GPU-Z idle
View attachment 339644

2nd screenshot shows GPU-Z 40 seconds into OCCT stability power test (CPU + GPU:3D)
View attachment 339645

3rd screenshot OCCT stability power test
View attachment 339646
same reco, install previous driver which was ok and use it as soon as newer than this buggy driver is released. I have had same issues with both amd and nvi cards.
 
PerfCap reason THRM at surprisingly low clock and power.

Downgrade the driver and see if the problem goes away
I've tried different versions. First uninstalled with DDU in safe mode.

1 time I got OCCT to start a power benchmark where the GPU went full speed, but I got a black screen after 5 seconds and laptop went off. After reboot GPU stays stuck in 210 MHz.
With other drivers it just stays stuck in 210 MHz. No change.

CPU at the moment is not stuck at 0.39GHz. Just performing normal. Only the GPU does not work well. Any other ideas?
 
Is the game being rendered in igpu or nvidia gpu?
 
Is the game being rendered in igpu or nvidia gpu?
Nvidia GPU. I actually disabled the igpu a while ago to exclude any issues. It was working fine then. I did enable it yesterday to check if it would make any difference, but with or without the problem persists.
 
i have somewhat a similar laptop 11800h and 3070.. i havent updated the driver... i will update tonight and install cyberpunk 2077 and see

i have installed cyberpunk 2077 and the latest driver and i didnt use ddu or anything, just straight install over... but i dont have geforce experience or app installed
just the driver... and cyberpunk runs well no drops in frames... spikes etc..

possibly there is window and nvidia driver's registry issue but I am not certain, only solution I can think of is backup and format and install fresh but use that as last option
maybe other members have other solutions
 
Last edited:
i have somewhat a similar laptop 11800h and 3070.. i havent updated the driver... i will update tonight and install cyberpunk 2077 and see
I've been having this issue for half a year now. It's a custom made Tongfang GM7AG7P notebook. It all worked great for a year, but I noticed performance issues with games like low FPS. Ever since it's been going downhill. Only a fresh install of Windows would get rid of the issue. Until Windows would force an update or games would tell me the drivers are outdated and should be updated.

GPU would get stuck in 210 MHz and CPU often gets stuck in 0.39 GHz. Unplugging and plugging the PSU in would trigger CPU to get out of 0.39 GHz state. I've tried all tips I've found so far, but nothing really solves the issue.

My supplier says I'm the only one with this issue and they can't take back the laptop, because a reinstall of Windows would fix the issue. Last but not least, I bought it in The Netherlands, but I live in Asia. So returning it will be quite a hassle.

--

Update while this post has not been approved yet. I've just enabled Allow access to the GPU performance counters to all users in the NVIDIA Control Panel under Developer > Manage GPU Performance Counters. The GPU immediately got out of 210 MHz and acted like normal. I started Cyberpunk 2077 and got good FPS, but after 5 minutes the screen went black and laptop shut down. Any ideas?
 
Last edited:
GPU-Z is accusing both power and thermal limits, so my best guess is that the driver is not working well with your laptop's power management system for some reason. In Throttlestop try to disable BD PROCHOT, MIGHT work. This may sound stupid but as a step further try the driver release that's on its manufacturer's driver page, see what's up.

Have had no issues with my 2021 Dell G15 5515 (AMD Ryzen 5 5600H with the original 4 GB/128-bit version of RTX 3050) with the latest 551.61 Studio drivers. Even my GPU overclock is working just fine. It's my workhorse machine.
 
GPU-Z is accusing both power and thermal limits, so my best guess is that the driver is not working well with your laptop's power management system for some reason. In Throttlestop try to disable BD PROCHOT, MIGHT work. This may sound stupid but as a step further try the driver release that's on its manufacturer's driver page, see what's up.

Have had no issues with my 2021 Dell G15 5515 (AMD Ryzen 5 5600H with the original 4 GB/128-bit version of RTX 3050) with the latest 551.61 Studio drivers. Even my GPU overclock is working just fine. It's my workhorse machine.

I can't disable BD PROCHOT in Throttlestop. It's greyed out and I've read that from 12th Gen this is done on purpose.

It's a custom made TongFang notebook and I got the drivers from the supplier, but the GPU driver for the 3070 Ti card they're pointing towards NVIDIA drivers website.

1710921682058.png
 
I can't disable BD PROCHOT in Throttlestop. It's greyed out and I've read that from 12th Gen this is done on purpose.

It's a custom made TongFang notebook and I got the drivers from the supplier, but the GPU driver for the 3070 Ti card they're pointing towards NVIDIA drivers website.

View attachment 339828

When your laptop's CPU is stuck at 400 MHz and the GPU goes super slow, what does the Limits window in Throttlestop say?
 
Don't ignore @W1zzard 's post. Your card is thermally throttling under little-to-no load. That's not normal.
 
I've been having this issue for half a year now. It's a custom made Tongfang GM7AG7P notebook. It all worked great for a year, but I noticed performance issues with games like low FPS. Ever since it's been going downhill. Only a fresh install of Windows would get rid of the issue. Until Windows would force an update or games would tell me the drivers are outdated and should be updated.

GPU would get stuck in 210 MHz and CPU often gets stuck in 0.39 GHz. Unplugging and plugging the PSU in would trigger CPU to get out of 0.39 GHz state. I've tried all tips I've found so far, but nothing really solves the issue.

My supplier says I'm the only one with this issue and they can't take back the laptop, because a reinstall of Windows would fix the issue. Last but not least, I bought it in The Netherlands, but I live in Asia. So returning it will be quite a hassle.

--

Update while this post has not been approved yet. I've just enabled Allow access to the GPU performance counters to all users in the NVIDIA Control Panel under Developer > Manage GPU Performance Counters. The GPU immediately got out of 210 MHz and acted like normal. I started Cyberpunk 2077 and got good FPS, but after 5 minutes the screen went black and laptop shut down. Any ideas?

If you have had the issue for half a year and tried all these things, i would say do a clean windows install on the laptop and see if it plays nice after that.
 
When your laptop's CPU is stuck at 400 MHz and the GPU goes super slow, what does the Limits window in Throttlestop say?
This is what I get:
1710937153663.png


PerfCap reason THRM at surprisingly low clock and power.

Downgrade the driver and see if the problem goes away

I'm sorry I did not respond earlier. I've tried many versions already. Even versions that used to work before without any issue.

In the last 24 hours I've tried:
535.98
537.42
551.23

It was working fine for 1.5 month on 551.23. This was the version I used after a fresh Windows 10 install, but it started to slow down over the last week and lag and lower FPS came back in games. Only had a few times CPU got stuck at 400 MHz. This always goes away after taking out and plugging in the PSU.

If you have had the issue for half a year and tried all these things, i would say do a clean windows install on the laptop and see if it plays nice after that.
I've had the issue before and a fresh Windows install helped. But I've done 3 of them in a short period, which is not normal IMO. After a while Windows updates come out which can cause the issue to come back. Also over time the FPS performance is going down slowly. Cyberpunk was not running as smoothly as before (after fresh W10 install).
 
Don't ignore @W1zzard 's post. Your card is thermally throttling under little-to-no load. That's not normal.

The idea is that with the driver, the thermal protection is tripping automatically when a certain driver version is installed. 90C under OCCT for the CPU and the GPU doesn't even crack 70, thermals are likely fine (it shouldn't be tripping thermal protection at 90 degrees C, it's too low)

This is what I get:
View attachment 339848



I'm sorry I did not respond earlier. I've tried many versions already. Even versions that used to work before without any issue.

In the last 24 hours I've tried:
535.98
537.42
551.23

It was working fine for 1.5 month on 551.23. This was the version I used after a fresh Windows 10 install, but it started to slow down over the last week and lag and lower FPS came back in games. Only had a few times CPU got stuck at 400 MHz. This always goes away after taking out and plugging in the PSU.


I've had the issue before and a fresh Windows install helped. But I've done 3 of them in a short period, which is not normal IMO. After a while Windows updates come out which can cause the issue to come back. Also over time the FPS performance is going down slowly. Cyberpunk was not running as smoothly as before (after fresh W10 install).

I see. ThrottleStop is saying the same thing, that your CPU is throttling due to heat. The BD (bi-directional) signal then tells your GPU to "whoa whoa slow tf down let's keep ourselves in check here". The end result is the mess you find yourself in. Whether this is due to a defect in a thermal sensor or a real hotspot problem with your CPU, this is what's essentially happening. ThrottleStop gets its data directly from the CPU, so the next logical step would to have the laptop serviced. You should do that. Maybe a reapplication of the thermal paste in the CPU is in order.
 
Last edited:
I see. ThrottleStop is saying the same thing, that your CPU is throttling due to heat. The BD (bi-directional) signal then tells your GPU to "whoa whoa slow tf down let's keep ourselves in check here". The end result is the mess you find yourself in. Whether this is due to a defect in a thermal sensor or a real hotspot problem with your CPU, this is what's essentially happening. ThrottleStop gets its data directly from the CPU, so the next logical step would to have the laptop serviced. You should do that. Maybe a reapplication of the thermal paste in the CPU is in order.

Is this something that could be serviced locally? I'm based in Vietnam, but my laptop is coming out of The Netherlands. So it's not that easy to get it back there. Just wondering. Thanks.
 
The idea is that with the driver, the thermal protection is tripping automatically when a certain driver version is installed. 90C under OCCT for the CPU and the GPU doesn't even crack 70, thermals are likely fine (it shouldn't be tripping thermal protection at 90 degrees C, it's too low)
It also says it's power throttling, so maybe a power delivery issue. Really not sure, but that definitely looks wrong. And I doubt the cause is the driver.
On the other hand, gaming on a laptop... OP was asking for it.
 
It also says it's power throttling, so maybe a power delivery issue. Really not sure, but that definitely looks wrong. And I doubt the cause is the driver.
On the other hand, gaming on a laptop... OP was asking for it.

Laptops are fine, as long as you don't expect desktop-class performance out of it. I mean, people buy Steam Decks and ROG Allys don't they? :D

Something wrong isn't quite right with OP's PC. The CPU is certainly responsible for issuing the bi-directional throttle signal, so before anything else i'd have it repasted, check if the fans are good, and then ensure that they are running near or at full speed during gaming sessions
 
Laptops are fine, as long as you don't expect desktop-class performance out of it. I mean, people buy Steam Decks and ROG Allys don't they? :D

Something wrong isn't quite right with OP's PC. The CPU is certainly responsible for issuing the bi-directional throttle signal, so before anything else i'd have it repasted, check if the fans are good, and then ensure that they are running near or at full speed during gaming sessions
Laptops are definitely not fine imho. Not going any further than the lack of options in the BIOS, options that are commonplace on a regular PC, that right there is a deal breaker for me.
You want to diagnose a video card on a PC? Remove the card, place it in another PC (or place a different video card in your own PC), see if it fixes the problem. On laptops, that's a no-go. And I could go on.
 
I've had the issue before and a fresh Windows install helped. But I've done 3 of them in a short period, which is not normal IMO. After a while Windows updates come out which can cause the issue to come back. Also over time the FPS performance is going down slowly. Cyberpunk was not running as smoothly as before (after fresh W10 install).

That would suggest that it's a windows issue. Fresh windows install, do regedit to prevent windows from updating - see if it stays fine.
 
This sounds like the power bug Nvidia cards can suffer from which is a hardware fault of the power sensing circuit signaling incorrectly high power draw but it's strange that the value reported is not actually abnormally high.

Is the GPU stuck in this state all the time ? Honestly at this point I would check this on a fresh install of windows with an old driver, if it's still behaving in the same way then it's because of the aforementioned hardware defect which could be intermittent.
 
Is this something that could be serviced locally? I'm based in Vietnam, but my laptop is coming out of The Netherlands. So it's not that easy to get it back there. Just wondering. Thanks.

I have no idea, but if you do not know how to disassemble and reassemble your laptop yourself, don't attempt to do this. Chance things will go wrong is very high.

This sounds like the power bug Nvidia cards can suffer from which is a hardware fault of the power sensing circuit signaling incorrectly high power draw but it's strange that the value reported is not actually abnormally high.

Is the GPU stuck in this state all the time ? Honestly at this point I would check this on a fresh install of windows with an old driver, if it's still behaving in the same way then it's because of the aforementioned hardware defect which could be intermittent.

First I heard of this :eek:

Is this something like an issue with the shunts on a regular desktop card?
 
Is the GPU stuck in this state all the time ?
No, it's not. I just got an EC update from the supplier. Installed it and played Cyberpunk. Got up to 80-90 FPS during 5 minutes and after that it went back to 8 FPS and GPU got stuck in 210MHz until I quit the game.

That would suggest that it's a windows issue. Fresh windows install, do regedit to prevent windows from updating - see if it stays fine.

I honestly think this could work. But is this a long term solution? Not updating Windows and GPU drivers for the next 1 or 2 years?
 
I honestly think this could work. But is this a long term solution? Not updating Windows and GPU drivers for the next 1 or 2 years?

Unless you're doing high security clearance work on that pc, i wouldn't worry at all about not updating windows. You will obviously still need to update gpu drivers for game support for new games.
 
Firstly, it's a gaming laptop, which is a problematic combination; I suspect there has to be something wrong with the power delivery circuit for the nv GPU. It's probably gotten too hot too many times over the years and degraded (designed with not enough spare amperage so to speak).
 
Back
Top